Notable projects demonstrating the power of clouds on OpenStack
Walmart, China Mobile, the Large Hadron Collider, and even Blizzard Entertainment games all use OpenStack. However, they do it in their own way.
OpenStack is an open source cloud solution that is highly flexible and customizable. Therefore, the solution copes with both ordinary tasks and highly loaded configurations.
Although OpenStack is not talked about as actively as 5 years ago, it has not gone anywhere. On the contrary: more and more companies are looking at it. For example, we are in “Inferit Cloud” We built the infrastructure based on this solution.
Today we look back at four cloud projects made possible by the flexibility and scalability of OpenStack.
CERN (European Organization for Nuclear Research)
CERN's main project is the Large Hadron Collider. Collider experiments generate enormous amounts of data. Cloud platform on OpenStack plays decisive role in managing this data. Scientists also provide infrastructure and resources to thousands of researchers around the world.
Private cloud by 2022 has grown up to 9,000 physical nodes with 500,000 cores. This infrastructure supports many of the scientific, administrative and computational tasks required for the functioning of CERN.
CERN implemented the project Worldwide LHC Computing Grid (WLCG). This is an integrated network of more than 170 computing centers in 42 countries.
WLCG uses OpenStack to provide the computing power needed to analyze the enormous amount of data produced by the collider. The result is that thousands of scientists around the world have access to data in real time.
zOpenStack allows CERN to provision massive computing resources quickly and efficiently. Users can request virtual machines and storage through a dedicated web portal, significantly reducing the time required to deploy resources. According to calculations CERN, without a deployed cloud network, this process could take up to several months. Fortunately, the process is now automated and takes a few minutes.
By the way, scientists continue to use magnetic tapes for long-term data storage. On average, such a tape stores data for 30 years and is not very demanding in storage.
CERN's OpenStack deployment includes advanced features such as container orchestration using Kubernetes and bare metal provisioning using OpenStack Ironic.
CERN cloud infrastructure loads can be evaluate in real time on their website.
In 2022, CERN experts said they were preparing to upgrade the collider. This will lead to an increase in data processing and will require new computing power. Scientists plan to remain in the OpenStack ecosystem.
Walmart
The world's largest wholesale and retail trade network – supporter open source products. Business at different times used more than 60 OpenSource solutions.
Walmart created a network of 10,000 servers running OpenStack-based cloud infrastructure in distribution centers and more than 5,500 stores. At the same time, Walmart's OpenStack private cloud has more than 800,000 cores.
The main feature of Walmart is Galaxy's own tool. Its task is to verify cloud operations. If a problem is detected somewhere, Galaxy reports it. This way the tool helps maintain the smooth operation of the cloud infrastructure.
Before Walmart, American retail was dominated by a reactive approach, where companies reacted to a problem when it was discovered. Galaxy works differently: it constantly runs checks to confirm that all components of the cloud are working. Today, a proactive approach to inspections has become mainstream.
Despite its partnership with Microsoft Azure, Walmart continues to invest in OpenStack and use it for its needs. The company uses a multi-cloud infrastructure, which it calls triune network. It integrates public and private cloud and edge nodes, allowing Walmart to flexibly balance workloads and maintain operational continuity during periods of peak activity or downtime.
Using OpenStack and Kubernetes, Walmart organizes running containers on an extensive computing base. The infrastructure supports rapid updates and the introduction of new features, which significantly increases the flexibility of Walmart's digital platforms. Moving to a multi-cloud environment helps you manage costs with flexibility and avoid vendor lock-in.
Despite competition from large cloud service providers, Walmart nevertheless adheres to Open Source solutions. Not long ago the company announced about plans to expand the pool with a tool based on open source solutions. Clouds based on OpenStack occupy a key position in these plans.
China Mobile
Officially: the largest mobile operator in the world, serving about a billion subscribers.
China Mobile operates one of the world's largest network function virtualization (NFV) systems based on OpenStack. This extensive infrastructure manages more than 50,000 servers, supporting an extensive telecommunications network that serves more than 3 million base stations.
China Mobile's NFV network supports more than 465 million 5G users and 300 million IMS (IP Multimedia Subsystem) users.
To effectively manage such scale, China Mobile developed automated AUTO platform, which significantly reduces the time required to configure and test equipment. AUTO can configure all devices in a resource pool of over 1,000 physical nodes in just 20 minutes and complete testing in 80 minutes. Within a year, this platform helped China Mobile reduce the number of configuration failures and eliminate more than 15,000 problems.
China Mobile has implemented a CI-CT-CD (Continuous Integration, Test and Delivery) pipeline in collaboration with Ericsson to integrate its solutions. This pipeline provides automated deployment and testing of new software versions, increasing the efficiency and reliability of network cloud services.
It is important that China Mobile actively participates in the OpenStack community, sharing its experience in the field of automation and CI/CD. The company also participates in conferences and events dedicated to the development of open source solutions. The community is most interested in the company's experience in using NFV in telecommunications environments.
Now the company has shifted its focus to the development of artificial intelligence, and the chairman of the board of directors of China Mobile statedthat the company has “moved from cloud to computing.” True, the income of cloud destinations for the past year grew up by 83%, and in 2023 the customer base of 5G users grew by 42% – the network withstood the influx of new subscribers. This means that using OpenStack brings results and they definitely won’t give it up.
Blizzard Entertainment
“I received power that my father never dreamed of!” The head of network infrastructure at Blizzard Entertainment could probably say so. Still, 12,000 computing hosts and several clouds are no joke.
Do you or someone you know play Overwatch 2, Diablo IV and World of Warcraft? Please know: To support the infrastructure of these games, Blizzard Entertainment uses OpenStack.
Blizzard primarily values OpenStack's auto-scaling capabilities to manage high and variable online gaming workloads. The number of players on weekdays, weekends, and during major updates means three completely different loads on the infrastructure. Automatic scaling helps maintain performance and stability by allocating the resources needed to meet the demands of game servers.
One of the techniques that Blizzard uses in OpenStack is called NUMA pinning. This technique allows you to optimize the use of computer processing power (CPU) by game servers. Roughly speaking, NUMA makes sure that processors use their memory sections, and the OpenStack Nova component performs primary tasks. This approach ensures that each virtual machine will be located in its own memory zone, without intersecting with the zone of other VMs.
Blizzard Company decided several challenges associated with scaling your gaming infrastructure. One such problem is solved with the help of a message broker – RabbitMQ, which helps different parts of the system communicate with each other, similar to how mail delivers messages.
Another major challenge was scaling Neutron, the networking component of OpenStack. Blizzard encountered with operational issues due to co-location of services on controller hosts.
Typically in an OpenStack environment, host controllers are responsible for managing and orchestrating various cloud services, including networking (Neutron) and computing (Nova). When multiple services run on the same hardware or virtual machines, they compete for resources. Under undulating loads, this competition can overload the system and reduce performance. Blizzard solved this problem by moving Neutron RPC employees to virtual machines, which reduced the load on the control plane to 75% during normal work.
Blizzard's approach to OpenStack involves constant tuning and scaling to maintain infrastructure efficiency and reliability. This strategy allows them to keep online games running smoothly, minimize glitches, and ensure performance.
Conclusion
From 2020 to 2021, the number of cores. managed by OpenStack, has grown by 66%.
OpenStack continues report about the emergence of new projects, tools and components. OpenStack has almost unlimited potential in terms of customization, so the emergence of new large-scale cases is a matter of time.
At the same time, there are many myths around OpenStack. For example, it is generally accepted that OpenStack is deployed manually, which takes a lot of time to fix errors. In fact, there is a class of tools that help in deploying OpenStack and its individual components.
It’s interesting that until recently it was fashionable to “bury” OpenStack, especially in the foreign segment of the Internet. He just used to be on the cutting edge of technology. Now OpenStack has settled in the bowels of data centers, serving as the basis for Kubernetes, AI/ML stack and telecommunications NFV.
If you've heard of other notable and well-known OpenStack projects, please share them in the comments.
